Robot behavior control based on current and predictive internal, external condition and states with levels of activations

• US 7,853,357 B2
• Filed: 03/09/2004
• Issued: 12/14/2010
• Est. Priority Date: 03/11/2003
• Status: Expired due to Fees

• Current Assignee: Sony Corporation (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.)
• Inventors: Takagi, Tsuyoshi, Sawada, Tsutomu, Fujita, Masahiro
• Primary Examiner(s): Black, Thomas G
• Assistant Examiner(s): Kong, Sze-Hon
• Application Number:
  US10/547,922
• Publication Number:
  US 20060184273A1
• Time in Patent Office: 2,471 Days
• Field of Search: 700 29- 31, 700/245, 700/250, 706/14, 706/15, 706/21, 706/23, 318/567, 318/568.1, 318/568.12
• US Class Current: 700/250
• CPC Class Codes: B25J 13/00 Controls for manipulators programme controls B25J9/16; control in general G05
• CPC Class Codes: G06N 3/008 based on physical entities controlled by simulated intelligence so as to replicate intelligent life forms, e.g. based on robots replicating pets or humans in their appearance or behaviour

Abstract

In a robot device, an action selecting/control system includes a plurality of elementary action modules each of which outputs an action when selected. An activation level calculation unit calculates an activation level AL of each elementary action on the basis of information from an internal-state manager and external-stimulus recognition unit and with reference to a data base. An action selector selects an elementary action whose activation level AL is highest as an action to be implemented. Each action is associated with a predetermined internal state and external stimulus. The activation level calculation unit calculates an activation level AL of each action on the basis of a predicted satisfaction level variation based on the level of an instinct for an action corresponding to an input internal state and a predicted internal-state variation predictable based on an input external stimulus.

Claims

1. A robot device which autonomously selects and implements an action on the basis of its own internal state and an external stimulus, the robot device comprising:

a plurality of action-stating modules each having an action stated therein;

an activation level calculating means for calculating the activation level of each of the actions stated in the plurality of action-stating modules from the internal state and external stimulus; and

an action selecting means for selecting one of the action-stating modules on the basis of the calculated activation level and performing the action stated in the selected action-stating module,

wherein the action stated in each of the action-stating modules are associated with a predetermined internal state and predetermined external stimulus;

wherein the activation level calculating means determines (a) an instinct level for an action associated with a current internal state, (b) a satisfaction level based on the current internal state, (c) a predicted internal state based on current external stimuli, the predicted internal state indicating a possible internal state as a result of implementing an action, (d) a predicted satisfaction level variation based upon the predicted internal state and (e) releasing state based upon the satisfaction level and the predicted satisfaction level variation,

wherein the activation level is calculated for each action on the basis of the instinct level and the releasing state.

2. The device according to claim 1, wherein the activation level calculating means calculates the activation level of each action from the instinct level determined from the current internal state, satisfaction level determined from the current internal state and the predicted satisfaction level variation.

3. The device according to claim 1, wherein the activation level calculating means has an activation level calculation data base in which input external stimulus and predicted internal-state variation calculates the activation level with reference to the activation level calculation data base.

4. The device according to claim 3, wherein the activation level calculation data base has the predicted internal-state variation associated with the value of the external stimulus.

5. The device according to claim 4, wherein in case a value not in the activation level calculation data base is supplied, the activation level calculating means calculates a predicted internal-state variation by the linear interpolation using a linear model.

6. A robot action controlling method for use in a robot device to control the robot device to autonomously select and implement an action on the basis of its own internal state and an external stimulus, the method comprising:

an activation level calculating step of calculating the activation level of each of the actions stated in a plurality of action-stating modules from the internal state and external stimulus; and

an action selecting step of selecting one of the action-stating modules on the basis of the calculated activation level and performing the action stated in the selected action-stating module, the action stated in each of the action-stating modules being associated with a predetermined internal state and external stimulus; and

in the activation level calculating step, determining (a) an instinct level for an action associated with a current internal state, (b) a satisfaction level based on the current internal state, (c) a predicted internal state base on current external stimuli, the predicted internal state indicating possible internal state as a result of implementing action, (d) a predicted satisfaction level variation based upon the predicted internal state and (e) releasing state based upon the satisfaction level and the predicted satisfaction level variation, and wherein the activation level for each action being calculated from the instinct level and the releasing state.

7. The method according to claim 6, wherein in the activation level calculating step, the activation level of each action is calculated from the instinct level determined from the current internal state, satisfaction level determined from the current internal state and the predicted satisfaction level variation.

8. The method according to claim 6, wherein in the activation level calculating step, the activation level is calculated with reference to an activation level calculation data base in which input external stimulus and predicted internal-state variation are associated with each other.

9. The method according to claim 8, wherein the activation level calculation data base has the predicted internal-state variation associated with the value of the external stimulus.

10. The method according to claim 9, wherein in case a value not in the activation level calculation data base is supplied, a predicted internal-state variation is calculated by the linear interpolation using a linear model in the activation level calculating step.

11. A non-transitory computer-readable medium storing a program to have a computer control a robot device to autonomously select and implement an action on the basis of its own internal state and an external stimulus, the program comprising:

an activation level calculating step of calculating the activation level of each of the actions stated in a plurality of action-stating modules from the internal state and external stimulus; and

an action selecting step of selecting one of the action-stating modules on the basis of the calculated activation level and performing the action stated in the selected action-stating module, the action stated in each of the action-stating modules being associated with a predetermined internal state and external stimulus; and

in the activation level calculating step, determining (a) an instinct level for an action associated with a current internal state, (b) a satisfaction level based on the current internal state,(c) a predicted internal state based on current external stimuli the predicted internal state indicating possible internal state as a result of implementing an action, (d) a predicted satisfaction level variation based upon the predicted internal state and (e) releasing state based upon the satisfaction level and the predicted satisfaction level variation, and wherein the activation level for each action being calculated from the instinct level and the releasing state.